Escolar Documentos
Profissional Documentos
Cultura Documentos
org
by 201.250.119.130 on 02/28/13. For personal use only.
ANNUAL
REVIEWS
Further
Quick links to online content
Ann. Rev. Anthropol. 1986. 15:1-24
Copyright © 1986 by Annual Reviews 1nc. All rights reserved
ON BEING A LINGUISTIC
ANTHROPOLOGIST
Annu. Rev. Anthropol. 1986.15:1-25. Downloaded from www.annualreviews.org
Joseph H. Greenberg
by 201.250.119.130 on 02/28/13. For personal use only.
Like those who have preceded me in wntmg Overviews for the Annual
Review of Anthropology, I became an anthropologist through a series of
accidents. It was still true during my undergraduate years at Columbia
(1932-36) that anthropology programs existed only at the graduate level, and
even then at only a few universities. The central figure was, of course, Franz
Boas at Columbia, and anthropology programs at other universities were, to
my knowledge, all founded by students of Boas . Anthropology during that
period had a very important component in museums and governmental in
stitutions such as the Bureau of American Ethnology, a role which has
diminished proportionally during my life. Because of Boas' stature in the field
and its chronological priority, the Columbia program during the period that I
was an undergraduate there had an indisputable primacy in the country.
As an undergraduate, however, I was completely unaware of all this until
the summer of 1935, just before I entered my senior year. During that summer
I chanced to read a volume edited by Calverton, The Making of Man (5),
available in the then popular and easily affordable Modem Library series. My
curiosity was aroused, and I noticed in the catalogue that there was a course in
general anthropology (actually the only undergraduate course offered) . It was
taught by Alexander Lesser, himself a student of Boas .
My interests up to that time seemed, at least on the surface, to be quite
different. My main field of study had been language which fascinated me
from an early age. During my high school years and even a bit earlier I had
acquired the habit of studying languages independently by reading grammars
and texts.
0084-6570/86/1015-0001$02.00
2 GREENBERG
they were the Ajax and and the Oedipus at Colonus. The Greek text was on
one side and the English translation on the other. I tried to analyze the Greek
texts with the help of the Abridged Oxford Dictionary of English which we
had at home. In this dictionary all English words derived from Greek were
given in the Greek alphabet. This allowed me to decipher the alphabet and
also provided me with a fair-sized vocabulary.
When I acquired the grammar, it was indeed a revelation. The varying
forms of the noun and verb which I had puzzled over now became clear.
There were cases, genders, tenses, and other categories that I had never heard
of before but some of which I would soon encounter in my study of Latin and
German. However, I never did find a dictionary so it was still very hard
going.
Later, when I attended Columbia, I took courses in Latin and Greek. In
addition, I had begun to teach myself Classical Arabic even in high school and
couldn't help noticing how similar it was to the Hebrew I had studied in
Talmud Torah. The reason for this similarity, of course, eluded me.
lt was the glory of Columbia in those days that there were scholars in such
languages as Arabic, Akkadian, and the Slavic languages who were members
of Columbia's teaching staff and who listed language courses in the catalog
which almost no one ever took. As a result, they were free to devote all their
energies to research. I suppose I was a sort of nuisance in that I took these
courses, invariably being the only student in the class.
By my junior year I had become aware of comparative linguistics and took
a course with Louis Gray, which as far as I can recall was essentially a course
in comparative Indo-European. Given these interests, had it not been for the
complete absence of academic jobs in such subjects during the depression, I
would quite naturally have become a Classical or Semitic scholar, or perhaps
a specialist in comparative Indo-European. The Chairman of Classics had
LINGUISTIC ANTHROPOLOGY 3
indeed informed me that there was no future for a scholar with an interest in
the Classics.
The events that changed the course of my life occurred in 1936-37. As I
mentioned earlier, my interest in anthropology had been aroused by the
reading of Calverton's book, and as a result, I enrolled in Lesser's course in
undergraduate anthropology. I noted also that there was a graduate course
given by Boas in American Indian languages, and I was given permission to
audit it. Actually, I fully participated in the course with perhaps five or six
graduate students. At about that time I discovered in the Columbia library the
"Handbook of American Indian Languages," edited with an introduction by
Franz Boas (4). I read all the grammars in the volumes. They seemed strange
Annu. Rev. Anthropol. 1986.15:1-25. Downloaded from www.annualreviews.org
and fascinating in their differences from each other and from the Indo
European and Semitic languages that I had studied.
All this occurred during my senior year, while the problem of what to do
by 201.250.119.130 on 02/28/13. For personal use only.
however, become acquainted with Werner Leopold, the great pioneer in the
field of child language acquisition.
In those days Yale was the great center for linguistics. As far as I know, it
was the only university with a separate linguistics department. The usual thing
in those days was for those interested in linguistics to participate in the
linguistics clubs that existed at every major university and were the common
meeting ground for faculty and students from English and various foreign
language departments as well as anthropologists. Even Leonard Bloomfield,
the acknowledged leader of linguistics in the United States, in the list of
founding members of the Linguistic Society of America published in Lan
guage in 1925, listed his affiliations as German and Linguistics. He had
originally been a member of a German department at Ohio State before
coming to Yale.
Because of my interest in linguistics and its virtual absence from North
western, Herskovits encouraged my plan to spend my second year of graduate
study at Yale. There I took courses from Leslie Spier and Robert Lowie . As
far as I can remember, I took only one course in linguistics and am no longer
certain whether I was a registered student. It was Bloomfield's course on
Comparative Indo-European. I did become acquainted with other linguists at
Yale, notably the Hittitologist and Indo-Europeanist Edgar Sturtevant and the
S anskritist Franklin Edgerton.
After the year at Yale, I undertook fieldwork in Northern Nigeria among
the Hausa. My topic was the comparison of the Muslim Hausa religion with
that of one of the very few communities of still surviving non-Moslem Hausa
(the Maguzawa). I learned Hausa and used it in my fieldwork and sub
sequently wrote a few articles about it. The importance of this experience was
not confined to Hausa. It aroused in me an interest in the very controversial
problems of the time regarding the classification of African languages in
LINGUISTIC ANTHROPOLOGY 5
which the genetic posItIOn of Hausa played an important role. This led
ultimately to my general work on the historical classification of African
languages (9).
In my next year of graduate work, under relentless pressure from Hers
kovits, which I appreciate in retrospect, I finished my dissertation, which was
subsequently published in book form under the title The Influence of Islam on
a Sudanese Religion (7).
After I received my doctorate there were just no academic jobs. I wished to
continue with linguistics and especially to work with Edward Sapir who by
,
that time was at Yale. I therefore applied for and received an SSRC postdoc
toral grant that once more took me to Yale. I heard there that Sapir was ill and
Annu. Rev. Anthropol. 1986.15:1-25. Downloaded from www.annualreviews.org
was living at home in New York. I felt very shy about meeting him and taking
up his time under the circumstances. As a result, I never made his acquain
tance, but I did audit courses with Bloch, Trager, and Whorf. In spite of my
by 201.250.119.130 on 02/28/13. For personal use only.
earlier contacts with Bloomfield, this was my first real acquaintance with the
sort of structural linguistics then dominant in the United States.
In the absence of any prospect for academic employment, I almost felt
relieved when in early 1940 I was drafted into the Army. I was supposed to
serve for one year, but in my heart of hearts I knew better. I actually ended up
serving for about five years, including a period overseas in North Africa and
Italy. When I was mustered out in 1945, there was a pentup demand for new
academic appointments because of the virtual freeze in wartime and the great
influx of veterans under the GI Bill of Rights.
I soon received a position at the University of Minnesota, where I stayed
during academic 1946--47. I then went to Columbia, where I remained until
my move to Stanford in 1962, where I spent the rest of my academic career.
From the preceding account it is clear that for me a professional career in
anthropology was essentially a way of practicing linguistics at a time when
even graduate programs in linguistics were almost nonexistent. Had I been
born 40 years later, I would in all probability have chosen to carry out my
undergraduate and graduate work in linguistics pr ograms.
In retrospect, however, I believe that the series of accidents that brought me
into anthropology was a fortunate one. It was because of the specific an
thropological training I had in African studies that my first major project had
to do with the historical classification of African languages, a project which
also gave me a wide acquaintance with a whole series of language structures
that were different from Indo-European, the only family of languages in
which I had any comparative training. More broadly, it made me aware of the
social and cultural dimensions of language in a far more vivid and direct way
than would have been possible in a purely linguistic program. Beyond that,
my first acquaintance with Lowie and his interest in the history of anthropolo
gy helped focus my rather diffuse interest in the history of ideas into a con-
6 GREENBERG
POSltlvlsm and the great work that laid the foundation of much of this
approach, namely the Principia Mathematica of Whitehead and Russell (20).
Sometime during my period at Yale in 1937-38, I remember that I was
talking to Bloomfield in his office, when he walked over to his bookshelves
and handed me a copy of Carnap's Logische Syntax der Sprache (6) (at that
time it was only available in the German original). He suggested that I read it.
I can still remember him saying, "A fellow might get a lot out of reading this
but, on the other hand, he might spend a lot of time and effort without it being
of any real value to him as a linguist." He evidently felt that at his age it would
be inadvisable to attempt it. I did read it and, of course, found it hard going
Annu. Rev. Anthropol. 1986.15:1-25. Downloaded from www.annualreviews.org
tion, then one could be assured that if the primitive propositions were true, it
was guaranteed that anything deduced from them would be true because of the
formal and purely mechanical nature of deduction in the calculus. Still, the
introduction of coordinating definitions with their semantic reference to things
in the world outside of the purely formal calculus did seem an inherently
,/
inexact procedure.
I may add that right from the beginning I was not convinced by the
so-called empiricist criterion of meaning which stated in its simplest form
(later subject to much revision) that the meaning of a statement was its method
of verification. If one could not state how it could be verified, then the
statement itself was meaningless.
At any rate, I felt that these were philosophical questions and that my own
primary interest was in language. However, I did find that there were two
valuable byproducts from the logical positivist approach and particularly from
the study of the Principia Mathematica that had preceded it historically. In an
axiomatic system, each step in reasoning was required to be justified by exact
methods of deduction, either from the primitive propositions of the system or
from statements which already had been deduced from these by the same
exact methods of deduction. Of course, the axiomatic method of geometry
based on Euclid had a basic similarity to that of positivistic axiomatics, but
my early exposure to Euclid did not have the same effect on me. This was
probably because I had encountered it at too early an age for it to affect me
strongly, but even more, I think, because the Principia and subsequent
systems of the logicist school gave explicit rules for deduction, and by not
giving the primitive propositions any meaning seemed to avoid the apparent
vagueness and arbitrariness of the Euclidean postulates.
In particular, the study of the Principia provided me with an intellectual
discipline which has ever since stood me in good stead. Further, it forced one
LINGUISTIC ANTHROPOLOGY 9
all, not only to positivist logicians but to American linguists of the period,
seemed to me to have a sufficient degree of organization, at least in some
favorable areas, that it could even be analyzed by means of an interpreted
axiomatic system. One such topic was kinship tenninoiogy, and one of my
earliest papers concerned the axiomatic analysis of kinship terms (8). It was
"universalistic" in one respect in that I examined a fairly large number of
tenninologies so that I could take into account all the categories that were
known to occur in actual systems. In regard to this, I was greatly assisted by
the pioneer paper of Kroeber (15), in which he discussed all the categories
that he knew to occur in the kinship systems of the languages of the world,
including such exotic ones as "state of connecting relative," e.g. the differ
ence in some Amerindian systems between daughter of my living brother and
daughter of my dead brother. As with axiomatic systems in general, the
notion here was to discover the basic and minimal number of tenns, relations,
and propositions by means of which all of the existing tenns in any language
could be defined once these primitive notions were, for any language, given
interpretations. I believe that this paper, along with several others by different
investigators at about this time, all had the basic notion that certain areas of
linguistic tenninology were systematically organized and were among the
precursors of the later ethnosemantics.
I sent a copy of my paper on the axiomatics of kinship to Carnap. He
replied and said that he was very pleased that the methods he had used could
in fact be applied to portions of natural language and that the logical positivist
approach could make some contribution to the social sciences.
I believe that even from the earliest period of my acquaintance with
American structural linguistics, I was at least dimly aware of certain aspects
of the then dominant theory which I felt were unsatisfactory. I think there
were three main sources of this rather vague feeling of dissatisfaction. One
was that I was by no means convinced of the efficacy of the procedures
10 GREENBERG
employed to provide definitions of the two basic units, the phoneme and the
morpheme. This was related to my second major source of disagreement
which had to do with the role of meaning and its study, the field of semantics,
within linguistics. Meaning seemed to me a central factor in language, given
that its very raison d'etre was communication, so that the study of both lexical
meaning and the meanings o f grammatical categories and relations must be an
integral part of linguistics.
However, in the structuralism of the 1940s and 1950s in the United States,
phonemes had a semantic function in that they were the minimal units which
distinguished meanings, so that, for example the minimal contrast between
Annu. Rev. Anthropol. 1986.15:1-25. Downloaded from www.annualreviews.org
bad and sad showed that band s were separate phon emes Similarly, the
.
morphemes were the minimal units which had a meaning as in the English
word play-er-s in which each of the three parts had a meaning and could not
by 201.250.119.130 on 02/28/13. For personal use only.
giving their translation, even though it was felt that on scientific grounds this
was not a very exact procedure.
A third important source of uneasiness with the then current theory per
tained to the position of historical and comparative linguistics. After all, a
large part of the formal training I had in linguistics was in these fields. Clearly
every lan guage was a product of historical evolution. Hence one kind of
reasonable explanation of many linguistic phenomena was the historical one.
Indeed, for the nineteenth century and the early part of the twentieth, to most
linguists this seemed to be the only kind of explanation. To me it still made
sense to explain linguistic phenomena historically. Yet the American
structuralism of the period emphasized the complete separateness of syn
chronic and diachronic linguistics. Historical considerations must never be
allowed to influence synchronic analysis. Moreover, although it seemed on
the surface that synchrony and diachrony were coequal branches of linguis
tics, it was clear that synchrony was at the center of the stage. When linguists
t alked of linguistic theory in this period, they simply meant the theory of
producing synchronic description. There really seemed to be no justification
for carrying on historical and comparative studies. They seemed to continue
LINGUISTIC ANTHROPOLOGY 11
by sheer force of habit and because certain people found them to be fascinat
ing topics.
When I came to Columbia in 1948, I found a very different sort of
linguistics in a dominant position. A large proportion of the linguists in the
New York area, most notably Andre Martinet and Roman Jakobson at Colum
bia, were Europeans who followed a very different form of structuralism, that
of the Prague school. They formed the Linguistic Circle of New York and
published the joumal Word with which I soon became involved editorially.
The main reason they wanted me, as I was well aware, was that I was almost
the only available linguist in the area whose native language was English.
Annu. Rev. Anthropol. 1986.15:1-25. Downloaded from www.annualreviews.org
was that its methodology was very loose compared to the rigorous procedures
to which I had become accustomed in American structuralism.
Among the writings of Prague linguists that I got to know during my initial
period at Columbia was the fundamental monograph of Trubetskoy on
phonology (19). Here, for the first time I found a linguist looking at a large
number of languages and comparing their structures. Although on the surface
it did not lead to explicit generalizations, it did show how to compare the
phonological systems of different languages in spite of the numerous dif
ferences in phonetic detail. This was done by setting up a limited number of
binary features, e.g. voiced versus unvoiced. Prague analysis thus went, as it
were, beneath the phoneme level. In theory any phoneme in any language
could be defined as a "bundle" of simultaneous features, e.g. voiced versus
unvoiced, stop versus fricative, nasal versus non-nasal, etc. These feature
oppositions were universal not in the sense that all phonological systems
possessed them, but that all systems utilized some of a rather limited set of
universal feature oppositions. There was therefore a general vocabulary by
means of which the systems of different languages could be compared in spite
of the differences of phonetic details. Moreover, there was one other aspect of
Trubetskoy's analysis the significance of which I did not at that period
discern. In regard to vowel systems, Trubetskoy had developed a typology.
All systems fell into three classes named on the basis of the shape of the figure
produced using the usual diagrams with a back-front dimension and a second
dimension of vowel height. However, it was not clear what further con
clusions could be derived from such a typological classification.
Trubetskoy's analysis also incorporated one further fundamental Prague
notion, namely that the binary oppositions each involved a hierarchy. Of the
members of each opposition, one of the members was, in a certain sense,
preferred over the other. This preferred member was called the unmarked
while the subordinate one was called the marked. The term marked was used
12 GREENBERG
ries by the 1930s; e.g. the singular was unmarked in relation to the plural. The
parallel with phonology was particularly close in regard to the property of
marking itself. For example, in English the marked category of the plural has
an overt mark, usually s, while the singular is expressed merely by its
absence. An additional and very basic property of the unmarked category was
its ambiguity in that it might, under certain circumstances, stand for the
category as a whole rather than as specific member of an opposition. This was
called neutralization. A semantic example is the unmarked status of "long" as
against short. The term "length" derived from "long" represents the entire
category.
A further important step had been taken by Jakobson. He tried to show that
there was hierarchy not merely within a single feature opposition, but the
features themselves were hierarchically organized (13). Some were more
basic than others. Most importantly, there were implicational relationships;
the more basic implied the less basic but not vice versa. Jakobson had
proceeded on a grand scale. The implicational hierarchy appeared not only in
adult normal language, but also in child language acquisition and language
loss in aphasia. In child language, learning the more basic was acquired
before the less basic. In aphasia, in mirror-image fashion the less basic was
lost before the more basic. Thus in all three situations the same implicational
relationships held in that the less basic was never found unless the more basic
was present. These are exciting ideas. Numerous exceptions and quali
fications have been found, but they helped guide research by generating
specific hypotheses and they still remain a fundamental insight.
In yet another respect, the Prague approach differed from that of American
linguistics. The separation of synchrony and diachrony was not quite as strict.
Although the center of interest was synchrony, certain members of the school
talked about dynamic synchrony. At any given synchronic stage, there were
some characteristics which were on the verge of extinction or were literary
LINGUISTIC ANTHROPOLOGY 13
reminiscences while others were recent innovations, or even old features that
had taken on new life and were in the process of spreading.
As can be seen, this view of language as undergoing, at any stage, dynamic
change also suggests that it is not a homogeneous entity. There are differences
of levels of style along with social and regional differences within the
population which are reflected in linguistic differences. If it should tum out
that this variation within the speech community is not random but can be the
subject of systematic study, we have one of the foundations for the modem
field of sociolinguistics. In fact, one of the prime movers in the development
of this area of study was William Labov, a student of the Uriel Weinrich,
Annu. Rev. Anthropol. 1986.15:1-25. Downloaded from www.annualreviews.org
himself a student of Martinet, a leading member of the Prague school who had
taught at Columbia.
However, in spite of my doubts regarding American structuralism and my
growing acquaintance with the Prague approach, I was still essentially loyal to
by 201.250.119.130 on 02/28/13. For personal use only.
the American school during the early 1950s. To me the Prague approach
seemed vague and inexact in comparison to that of American structuralism.
It was in 1953, and once more through the intervention of the Social
Science Research Council, that 1 underwent an experience that had far
reaching consequences for my subsequent work. The Council had in the
previous year formed a Committee on Linguistics and Psychology of which I
was not at that time a member. It held a two-week seminar at Cornell. The
results seemed sufficiently encouraging for the Committee to plan a summer
seminar in conjunction with the Summer Institute of Linguistics at the Univer
sity of Indiana in 1953.
The seminar was based on the assumption that three areas-linguistics,
psychology, and information theory-had developed to the extent that in
teraction among specialists in these fields would be fruitful and might perhaps
help toward their ultimate integration. The seminar itself consisted of a small
number of faculty members and a few graduate students from the three fields.
The first step was to acquire some elementary acquaintance with each
other's fields. It fell to me to give the basic initial exposition of linguistics in
several two-hour sessions. I described with some pride, though not without
inner misgivings, the rigorous method by which a linguist confronted with a
corpus, as it was then called, of utterances from a language could discover the
basic units of phonology and grammar, the phoneme and the morpheme
respectively, as well as the rules concerning permissible use of these elements
on both levels.
When I had finished, Cornelius Osgood, one of the psychologist members
of the seminar, asked me a question which was to haunt me thenceforth and
helped determine the direction of much of my future work. I cannot now
recall his exact words, but they were approximately the following. "You have
described a very impressive procedure for analyzing any language into its
14 GREENBERG
basic units. However, if you could tell me something that was true about all
languages, that would be of interest to psychologists."
It was this remark that brought home to me the realization that all of
contemporary American linguistics consisted of elaborate but essentially
descriptive procedures. One could go on making analyses of language after
language. In each case the concrete meaningful elements and the grammatical
categories would tum out to be different. The only thing that was universal
was the procedure for arriving at these differing results. What one would do
with these various grammars after writing them was not clear.
The topic of language universals which was broached in this way turned out
Annu. Rev. Anthropol. 1986.15:1-25. Downloaded from www.annualreviews.org
Language (11).
After the summer seminar at Bloomington, I continued my interest in the
subject of language universals, but there was one basically disturbing ques
tion. Assuming that it was important to discover generalizations which were
valid for all languages, would not such statements be few in number and on
the whole quite banal? Examples would be that all languages had nouns and
verbs (although some linguists denied even that) or that aU languages had
sound systems and distinguished between phonetic vowels and consonants.
It was at this point that another linguistic topic I had just begun to work on
became relevant, that of typology. When linguists mentioned typology during
this period, what they usually had in mind was the traditional nineteenth
century classification of languages into isolating, agglutinative, and inflective
which was in general disrepute. The classification was inexact in that the
definitions of the types were never stated with sufficient clarity, and also
ethnocentric. It was assumed that the three types formed a sort of evolutionary
progression and that the highest stage, inflective, had only been reached by
the Indo-European and Semitic groups. However, since the more recent
Indo-European languages such as English and French were generally less
inflective than the older languages such as Latin, Greek, and Sanskrit, a
second basis of classification was introduced to account for this, namely that
between synthetic and analytic. The modern Indo-European languages were,
in general, analytic in that they "analyzed out" what in languages like Latin
would be fused in a single inflection. For example, Latin puero, the dative
singular of puer "boy," had an inflection that expressed in a single vowel the
categories of case (dative), number (singular), and gender (masculine) which
were separately expressed in analytic languages.
In the early 1950s, Voegelin in particular sought to revive interest in
typology. There were two reasons why I took the topic seriously. One was
LINGUISTIC ANTHROPOLOGY 15
developed quite often outside of Africa, e.g. in Southeast Asia and in the
many indigenous languages of Mexico, independent of its occurrence in
Africa.
The research on African linguistic classification had been carried out
mainly in the period 1949-1950. In regard to typology it had two main effects
on me. The first was the realization that typology was a broader subject than
the traditional nineteenth century typology. This notion was reinforced when I
became acquainted with Trubetskoy's phonological typology. Moreover,
although typological criteria were irrelevant to historical classification they
seemed too important not to have some significance, perhaps of a different
sort.
The second reason for my interest in typology was simply that Sapir, whom
I admired greatly, had made it a central topic in his seminal book Language
(17). In his book Sapir had sought to reanalyze the criteria employed in the
traditional ninteenth century typology into a number of different and in
dependent dimensions, as well as to rid it of its ethnocentrism. In 1954, the
year after the summer seminar at Indiana, I tried to formulate a reanalysis of
the traditional typology not unlike that of Sapir but with less reliance on
intuition both for the definitions of typological dimensions and for the assign
ment of languages to particular typological classes (10). In fact, I quantified
the typology so that instead of stating that a language was highly synthetic, I
described a procedure which might lead to the calculation of a synthetic index
of, say .94, whereas a language which was far less synthetic might have an
index of synthesis with the value .17.
In the period between 1954 and my residence as a Fellow at the Behavioral
Sciences Center in 1958-59, I gradually came to the realization that a
systematic treatment of typology in its broader sense might, in fact, answer
the question that had troubled me about the meagerness and relative triteness
of statements that were simply true of all languages (what I came to call
16 GREENBERG
i ly rushes in (hence the term implosive) , but the inward direction of the air
stream is quickly reversed as the lung air rushes out after the glottal closure is
released.
by 201.250.119.130 on 02/28/13. For personal use only.
languages with back but not front implosives, 3. languages with front but not
back implosives, 4. languages without either front or back implosives .
Of these four types, type 3, with front but not back implosives, is the only
one not known to occur. This nonrandom distribution of languages among
types can be restated as an implicational universal. The presence of back
implosives in any language implies the presence of front implosives, but not
vice versa. Moreover, back implosives show the usual characteristics of
marked categories.
An unrestricted universal such as that all languages have phonetic vowels
can then be considered the logically limiting case involving the simplest
possible typology, one with a single dimension and two values. In regard to
vowels there are two logically possible types, languages with vowels and
languages without vowels . The second type is not found in actual languages.
When I began to realize the significance of implication and marking
relations, I experienced what Germans have called the Aha-Erlebnis. So this
was what Jakobson and other members of the Prague school had been driving
at all this time!
In 1958-59 I was a fellow at the Center for Advanced Study in the
Behavioral Sciences, Stanford, Calif. It had been planned that a number of
members of the Indiana summer seminar would be present. Among these
were myself and two of the psychologists, Cornelius Osgood and James J.
LINGUISTIC ANTHROPOLOGY 17
change in their point of view. For good or ill, this was, I think, the case with
me from about 1960. The investigation of typological characteristics of
language on a broad scale and of the regularities that can be derived from
them is obviously beyond the capabilities of a single researcher. During the
by 201.250.119.130 on 02/28/13. For personal use only.
body and all languages have additional meanings which derive from them that
are either synchronically obvious as in English 'in back of' or a survival of a
body part term in transferred meaning when the source meaning has become
obsolete.
The first extension, if it occurs, always seems to be to spatial relations and
only later, in some instances , to temporal and more abstract concepts . Once
more it is possible to formulate such relations as implications . However, in
diachronic universal implications there is an asymmetry not found in the
synchronic case. The later implies the earlier and not vice versa. Thus
Annu. Rev. Anthropol. 1986.15:1-25. Downloaded from www.annualreviews.org
temporal , and abstract terms which derive historically from such body part
terms . Hence no change in type is involved.
My other major concern besides typology and universals has been with
historical comparative linguistics in the traditional sense and with the genetic
classification of languages. Historical and genetic studies are related to the
study of language universals in a number of ways . In the study of diachronic
processual universals as described above , every individual case of develop
ment that enters into the comparison requires the application of the com
parative method in the traditional sense.
A second important link between typology/universals and historical linguis
tics concerns the problem of language sampling in synchronic studies . When
we assert, on the basis of a particular sample of languages , that a certain
linguistic property is related to another, it becomes relevant to investigate the
extent to which the properties in specific languages are independent, and by
this we mean historically independent.
This question is not reducible merely to whether the languages are related.
If all the languages of the world had a single origin so that they were all
ultimately related, the problem would still remain. The question in each case
has to do with the historical independence of particular linguistic traits .
Related languages often share typological characteristics which are not part of
their common inheritance but have developed independently since their sepa
ration (convergence).
For example, the basic word order verb-subject-object is found generally in
early Semitic languages and widely in other branches of Afroasiatic, the
larger stock to which Semitic is affiliated . Doubtless the occurrences of the
opposite, verb final order, subject-object-verb in both ancient Akkadian and
modern languages of the southern branch of Ethiopian Semitic , such as
Amharic, are historically independent developments in spite of the Semitic
affiliations of both.
LINGUISTIC ANTHROPOLOGY 19
The first step in the historical comparative method itself is genetic classifi
cation because without it we will not know which languages to compare . Here
again we are not talking merely about whether all the languages are related.
We wish to distinguish valid genetic units at whatever level. Thus the
Gennanic l anguages are a valid genetic unit in the sense that they are all more
closely related to each other than any of them is to any non-Gennanic
language. A group consisting of Swedish, Polish, and Albanian is not a valid
genetic unit and therefore not the basis for comparative historical study even
though all these languages are related, since they are all Indo-European.
Annu. Rev. Anthropol. 1986.15:1-25. Downloaded from www.annualreviews.org
The genetic classification of all the world' s languages has, of course, not
been completed. In this respect linguistics compares unfavorably with biOlogy
because in the latter field the task was already completed by Linnaeus in the
eighteenth century . Of course major modifications have since been made but
by 201.250.119.130 on 02/28/13. For personal use only.
in many fundamental respects it has held up. It is likely that the task is more
difficult in regard to language, but I believe we are now at the stage at whic h ,
i f it is feasible , it should b e possible to carry it out.
If this major task, which has been relegated to the periphery in contempo
rary linguistics, could be completed , we would have a tool of immense
importance for human cultural and biological history. Even the partial results
obtained up to now are obviously valuable. For example, the linguistic
comparative work and the reconstruction to a considerable extent of the
vocabulary of the Proto-Indo-European speech community connects in impor
tant ways with archeology and human genetics.
In describing how I became a linguistic anthropologist, I noted that al
though it resulted from a series of accidental events , given my interest in
language, it seemed a natural decision to enter anthropology . In American
anthropology, as contrasted with British social anthropology and the French
tradition, a synoptic view of man embracing both physical and cultural
aspects prevailed. In the so-called four fields approach , to which most of
American anthropology adheres at least in principle, anthropology consists of
four subfields: cultural anthropology , linguistics, archeology , and physical
anthropology.
It is obvious that this division has no consistent logical basis. Language is a
part, indeed a central part of culture; it is simply that the complexity of its
study made it a separate field and this separateness has continued and in
creased over the course of time. Archeology , once more , is essentially a part
of cultural anthropology but involves a set of techniques which also have
become greatly specialized over the course of time.
If we consider the four conventional subfields of anthropology , it also clear
that from the beginning the branches were not of equal importance . Cultural
anthropology was and is central, and although there are a few academic
departments with strong concentrations in fields other than cultural anthropol-
20 GREENBERG
ogy, many smaller departments have little or only token representation in the
other fields .
There have been two major developments since the period when I entered
anthropology which have tended toward the diminution of the role of linguis
tics within anthropology and of the position of linguistic anthropology in
relation to linguistics as a whole.
One has been the continual development of new specialties within cultural
anthropology (a glance at the Annual Review of Anthropology through the
years is enough to illustrate this). Arensberg ( 1) , in his overview in the first
Annu. Rev. Anthropol. 1986.15:1-25. Downloaded from www.annualreviews.org
issue of the Annual Review series, called attention to this, noting that ". . . the
progress of anthropology displays a proliferation and sprawl that threatens its
advance. " Some of the pioneers like Boas and Kroeber worked in several
branches of anthropology, which frequently included linguistics. When I was
by 201.250.119.130 on 02/28/13. For personal use only.
at Columbia, Kroeber taught there in the early 1950s after his retirement from
Berkeley. He was sufficiently interested in linguistics to go with me regularly
to the meetings of the Linguistic Circle of New York. Linguistics was still not
so esoteric that an interested outsider could not understand it. Among the
founding members of the Linguistic Society of America we find a number of
anthropologists who were not linguistic specialists. During the initial period
of my career, I wished to work both in linguistics and cultural anthropology
but soon found that this was no longer possible, at least in regard to research.
The development of new specialties has clearly diminished the proportional
role of linguistics within anthropology. This has been accompanied in the last
two decades by an explosive growth of linguistics as an independent disci
pline. Most universities, whatever their size, now have separate linguistics
departments , often of considerable size.
The result of these developments has been to question with increasing
insistency what, if any, distinctive role remains for linguistics within an
thropology. On the one hand, it is clear that as long as anthropology claims
human culture as its central domain, there are both practical and theoretical
reasons for it not to abandon linguistics. On the other hand, one may well ask
why a graduate student with a basic interest in language would choose to enter
an anthropological rather than a linguistic program.
I would maintain that there is still a role for linguistic anthropology. In
attempting to show this, I will outline two basic approaches to language. To
avoid misunderstanding it should be made clear that the division is not an
absolute one. Many individuals will be involved to some extent in both.
Moreover, on the theoretical plane each can and must fructify the other. Nor
do I claim the superior importance of one over the other. Again, although a
rough correlation with diSCiplinary lines will be evident, it is far from
absolute.
LINGUISTIC ANTHROPOLOGY 21
With all these qualifications in mind, we can still characterize two ways of
regarding language. One is language as a sociocultural institution and the
other as a system of signs that can be the subject of fonnal analysis. The term
formal is not here intended to exclude semantics, for which there is a place in
present-day formal approaches. I believe this will be clear in the sequel .
Since the notion of language as a human cultural institution will be im
mediately clear to anthropologists, nothing further will be said about it for the
present. The other way of looking at language, which is attractive in some
degree to all linguists , can be illustrated by a simple example.
Let us consider a restricted part of elementary arithmetic and consider in
Annu. Rev. Anthropol. 1986.15:1-25. Downloaded from www.annualreviews.org
what ways it resembles ordinary spoken language and in what ways it differs .
This "language" will contain only statements involving cardinal numbers ,
both positive and negative, and the operations of addition and subtraction. All
by 201.250.119.130 on 02/28/13. For personal use only.
but with good will on the part of the receiver of the message can be interpreted
by supplying a closed parenthesis after 2 .
I t i s obviously akin to language i n having a finite vocabulary; i t further
resembles natural language in that some items such as the cardinal numbers
are purely lexical while others such as the parentheses, which show that the
enclosed items belong together and express also the order of operations, are
much like inflectional markers or significant word order. They thus express
syntactic relations and seem to belong more to the internal mechanism of the
language.
Even some of the "concrete vocabulary" requires additional semantic rules
in certain contexts. For example, 13 in 13 6 + 7 is interpreted as 10 times 1
=
plus 3. Natural language also contains such complications. The closest paral
lel is the numerical system of certain languages like Chinese in which the
22 GREENBERG
much like phrases and clauses that are constituents of sentences in natural
languages. In 3 + (5 + 7) = 12, 5 is in closer syntactic relation to 7 than to 3
and (5 + 7) can be considered a sort of constituent within the expression as a
whole.
These and other parallels that might be pointed out suggest that "natural
languages" are a species within a larger class. The by now somewhat archaic
term "sign system" has been replaced in general use by "language," while
those spoken by human beings like English and French belong to a subclass
called "natural languages . " One talks now about "computer languages" as a
subset of languages alongside of natural languages. It is clear that this way of
looking at natural languages as but one subclass belonging to a more inclusive
category of language in a broader sense can be exceedingly fruitful in regard
to certain kinds of problems.
There are, of course, important differences between the restricted portion
of arithmetic we have been discussing and natural languages . Some of these
recur when we compare natural languages to other language systems and
some do not. However, there remains an important core of properties that are
unique to natural language . Some pertain to its inner structure and some to its
relation to its users and the speech setting in general .
For example, arithemetic lacks "indexical signs" such as pronouns and
demonstratives whose reference shifts on each individual occasion but have a
reasonably constant relation to the participants in the discourse, to the dis
course which has preceded it, and the physical setting in which it takes place.
Ordinary languages can of course also express general truths or falsehoods
that are independent of the momentary setting.
The lack of indexical signs in the language of arithmetic is related to the
kind of truth values of its statements which are eternally ("logically") true or
false. The true statements of mathematics thus consist entirely of tautologies,
LINGUISTIC ANTHROPOLOGY 23
but in individual more complex cases, the problem is precisely to show if they
are logically true or logically false. A mathematician friend of mine once cited
a de finition of mathematics as the study of interesting tautologies.
Ordinary languages have further numerous characteristics unrelated to their
structure which make them unique and fundamental. They have priority in the
history of humanity and of the individual . Spoken language existed long
before the first invention even of writing, with which it has peculiar and
important relations and a fortiori to other sign systems . Every individual
learns to speak before learning mathematics or computer languages , and the
priority is not merely chronological . We learn the numbers and even some of
Annu. Rev. Anthropol. 1986.15:1-25. Downloaded from www.annualreviews.org
their priority and centrality in non structural aspects to other systems with
which they share important commonalities of structure .
Anthropology, as the most comprehensive social discipline , will of necessi
ty continue to include the study of natural language as an essential part of its
task . This does not exclude either the role of linguistics or language de
partments in the study of natural languages. It is also compatible with the
more formal approaches that are so prominent in present-day linguistics.
Indeed, it is precisely the study of these similarities and differences that will
put natural language in a broader perspective, while in turn the comprehensive
typological and historical study of natural languages , the most complex
member of the genus, will help to shed light on the nature of communication
systems in general.
Literature Cited
aphasie und allgemeine lautgesetze. Up Harcourt & Brace. 258 pp.
psala: Almqvist & WiksclJ 1 8 . Shtemberg, L. I . 1 933. Gi/'aki, Drochi,
14. Joos, M . , ed. 1957. Readings in Lin Gol'di, Negidaltsi, Aini. Khabarovsk:
guistics: The Development of Descrip Dal'giz. 740 pp.
tive Linguistics in America since 1925. 19. Trubetskoy , N. S. 1939. Grundzuege
Washington, DC: Am. Counc. Learned der phonologie. Trav. cercle Linguist.
Soc. Prague 7 .
1 5 . Kroeber, A. L. 1 909. Classificatory sys 20. Whitehead, A . N . , Russell, B . 1 925-27.
tems' of relationship . J. R. Anthropol. Principia Mathematica, 3 vols. Cam
inst . 34 : 77-84 bridge: Cambridge Univ. Press. 2nd ed.
16. Loun sbUry . F. 1959. Language . Bienn. 2 1 . Working Papers on Language Univer
Rev. Anthropol. 1959: 1 85-209 sals. 1 969-76. Language Universals
1 7 . Sapir, E. 192 1 . Language, an Introduc Project, No. 1-20. Committee on lin
tion to the Study of Speech. New York: guistics, Stanford Univ. , Stanford,
Annu. Rev. Anthropol. 1986.15:1-25. Downloaded from www.annualreviews.org
Calif.
by 201.250.119.130 on 02/28/13. For personal use only.